Excluding data

[Jon Patrick]

    Date:       Thu, 30 Sep 1999 11:38:29 +0100 (BST)
    From:       Larry Trask <larryt at cogs.susx.ac.uk>

    On Mon, 27 Sep 1999, Jon Patrick wrote:

    > I'm concerned that the current extent of the thesis on early Basque
    > phonology is already based on the subset of data restricted by your
    > non-phonological criteria.

    But how can this be so?  How can my non-phonological criteria for
    selection affect the phonological properties of the words selected?
    Please give me a reply of substance, and not just vague, dark hints that
    something or other sinister is going on.

No dark hints - it is entirely possible that the criteria have been selected
so that
they only admit words that will provide the answer that you "believe" is true.

    Further: what *alternative* criteria do you propose for identifying the
    Basque words which are most likely to be native and ancient?  I've
    already put my criteria on the table.  What are yours?

MY approach is defined by a method rather than by criteria and produces a
study wider and lengthier in scope. I believe that it is necessary to take the
Azkue list of monomorphemes and for each meaning (not word) determine which
can been shown to be borrowed. The residual represents the data set of native
basque words. Any further evidence of morphological devleopment would need to
be included in the database.  I also propose that the analysis of all meanings
needs to presented in a database that contains justification for each decision
of categorisation. Such a database then has to become a public piece of
evidence for others to process as they see fit.   Only then can issues such as
this or that criteria for limiting the dataset in analysis  be fairly
appraised, as only then would one scholar be able to deal with the SAME data
set as others and evaluate their own pet theories in a direct and publicly
accountable display. That is, we all could actually evaluate the effect of
your criteria per se, or in competition with any other criteria anyone cared
to produce.

    > This thesis is the basis of many of your
    > comments as immediately below and I muse over the question which
    > should come first, the rules that declare a word's form to be
    > "curious" or the systematic and rigourous analysis of all the words.

    My observation that a particular word has a "curious" form has
    absolutely nothing to do with whether it goes into my list or not.
    In my view, the words <ke> `smoke' and <mutur> `snout, muzzle' have very
    curious forms indeed, but they meet my criteria and so must go into my
    initial list, whether I like it or not.  Once I've chosen my criteria, I
    have to stick to them.

    When I observe that a given word has a "curious" form, that happens
    because I've already done a good deal of work on Pre-Basque phonology,
    and I already have a pretty clear idea of what's going to emerge when
    the whole thing is done -- though, as I've already said, I do expect to
    find a few surprises waiting for me.  But I never exclude a word merely
    because I judge it to have a "curious" form.

Larry, that's precisely the worry, that your criteria are excluding entirely
valid words. Whilst I understand they seem perfectly sensible to you, others
don't share the same view because they are excluding and people are concerned
that excluded words are never actually available for them to scrutinise the
validity of excluding them. Now you can tell us to go and look up the words
ourselves but that is now excluding not only those words but us also from the
process as we don't know what specific words you have excluded.
hence my proposal above -one starts by showing everyone the comprehensive list
and telling them why EACH item is excluded.

Jon
______________________________________________________________
The meaning of your communication is the response you get